Orbital scale variations and timescales from the Arctic
OceanOrbital scale variations and timescales from the Arctic
Ocean
Heiko Palike,1 David J. A. Spofforth,1 Matthew O’Regan,2 and Jerome
Gattacceca3
Received 30 April 2007; revised 23 December 2007; accepted 8
January 2008; published 19 March 2008.
[1] We investigate geochemical measurements from the Arctic Coring
Expedition to derive sedimentation rates for selected intervals in
the early and middle Eocene. This analysis is performed by
quantifying cyclical variations in physical property data and
elemental concentrations derived with an X-ray fluorescence (XRF)
core scanner. Our results show that physical properties and
XRF-derived elemental concentrations are coherent and correlated to
each other. Changes in elemental concentrations occur on depth
scales from decimeters to meters and correspond to varying
contributions of detrital minerals and biogenic silica. We confirm
and refine sedimentation rates of the order of 10 to 25 m Ma1.
These rates were obtained independently through biostratigraphic
and magnetostratigraphic methods. We observe a strong imprint of
astronomically sourced cycles, particularly through climatic
precession. This observation allows us to test recently proposed
theoretical calculations of insolation.
Citation: Palike, H., D. J. A. Spofforth, M. O’Regan, and J.
Gattacceca (2008), Orbital scale variations and timescales from the
Arctic
Ocean, Paleoceanography, 23, PA1S10,
doi:10.1029/2007PA001490.
1. Introduction
[2] Accurate determinations of ages and sedimentation rates in
marine cores are an extremely important prerequisite to decipher
past geological, biological and climatological processes. The
determination of such rates typically involves multiple approaches,
ranging from direct radioisotopic dating [Kuiper et al., 2004] to
correlative comparison with existing dated records through
magnetostratigraphic or biostratigra- phy, or by comparison with
theoretically derived astronom- ical insolation curves [Laskar et
al., 1993; Laskar, 1999; Laskar et al., 2004], or through a
combination of the aforementioned [Lourens et al., 2005]. [3]
Advances in data acquisition and techniques have
been made recently and enabled the compilation of a highly accurate
and resolved geological timescale for the Neogene [Lourens et al.,
2005]. These timescales rely fundamentally on the Earth’s built-in
metronome. The metronome takes the form of astronomically driven
Milankovitch-type variations in Earth’s insolation. Changes in
Earth’s insolation quanti- ties are indirectly recorded in the
sedimentary archive after having passed through the filtering
system of Earth’s climate machine. Unlike radioisotopic dating
methods, the observation of astronomically driven sedimentary
cycles offers similar precision during the Paleogene as in the
Neogene. Potentially, relative dating errors can be as small as a
single astronomical cycle, e.g., 22, 41, 120 or 405 ka [Laskar,
1999].
[4] High-resolution age models are not only necessary to estimate
geological rates of processes, but can also be used to extract
information about the climate system and underlying astronomical
variations [Shackleton, 2000; Palike and Shackleton, 2000; Palike
et al., 2004; Billups et al., 2004; Wade and Palike, 2004; Raffi et
al., 2006; Palike et al., 2006a, 2006b]. Invoking astronomical
inso- lation cycles to explain ice ages was advanced by
Milankovitch [1941], who argued that summer insolation in high
northern latitudes is the controlling factor for ice sheet advance
and retreat. Milankovitch postulated that if summer insolation
values remain low enough, temper- atures are also low, and prevent
the significant melting of ice during summer. At present, the
majority of authors who use insolation calculations still use the
65N summer insolation as the forcing function, although
modifications to this theory have now been proposed [Huybers,
2006]. High-resolution age models generated from sediment cores at
extremely high latitudes, such as those recovered during Arctic
Coring Expedition (ACEX) [Backman et al., 2006; Moran et al.,
2006], offer the opportunity to test the sensitivity of local
climatic forcing as a function of latitude. [5] The analysis of
climatic cycles, responding to
insolation forcing, typically requires continuous, long and
high-resolution records. Additional age control from
biostratigraphy or magnetostratigraphy is required to pro- vide
anchoring points in the nearly repetitive pattern of Earth’s
insolation. This additional control is also neces- sary to
determine which of the three contributors to insolation variations
(climatic precession, obliquity, short and long eccentricity) are
dominant in any given mea- sured parameter. [6] In the case of the
ACEX record, these requirements
are only partially met, as stratigraphic sections below the top few
tens of meters are not fully recovered, and age control is mostly
provided by relatively low-resolution
PALEOCEANOGRAPHY, VOL. 23, PA1S10, doi:10.1029/2007PA001490,
2008
1National Oceanography Centre, University of Southampton, South-
ampton, UK.
2Graduate School of Oceanography, University of Rhode Island,
Narragansett, Rhode Island, USA.
3Geophysics and Planetology, CEREGE, CNRS, University of Aix-
Marseille 3, Aix-en-Provence, France.
Copyright 2008 by the American Geophysical Union.
0883-8305/08/2007PA001490
PA1S10 1 of 13
biostratigraphy. Thus, the same factors that make it difficult to
achieve a high resolution and an accurate age model in the ACEX
cores also hinder their cyclostratigraphic analy- sis.
Nevertheless, the physical properties measured from the ACEX cores
[Backman et al., 2006; Moran et al., 2006], as well as additional
high-resolution data (organic carbon contents [Stein et al., 2006],
X-ray fluorescence (XRF) core scanner elemental concentrations
[Backman et al., 2006; Spofforth et al., 2008], and biological
proxies [Sangiorgi et al., 2008b]), show a striking cyclical
pattern during certain intervals on decimeter to meter scales. We
exploit these cyclical variations in high-resolution meas- urements
(1) to estimate relative sedimentation rates through several time
slices where recovery was good, and (2) to derive a better
understanding of what, if any, cyclicity is recorded in the unusual
environments of the Arctic Ocean. Work by other authors has shown
that climatic forcing can be represented through proxy records in
the Arctic through several mechanisms, including varia- tions of
sea ice extent and intensity [Krylov et al., 2008], variations in
productivity and anoxic or suboxic preservation [Stein et al.,
2006; Sangiorgi et al., 2008a], variations in terrigenous input
[Spofforth et al., 2008], variations in freshwater supply
[Sangiorgi et al., 2008a], and a close relationship between
sediment properties (color, density, grain size) with climatic
state [O’Regan et al., 2008].
2. Materials and Methods
2.1. Physical Property Measurements
[7] As part of our analysis, we made use of the following physical
property parameters measured during the onshore and offshore phase
of ACEX [Backman et al., 2006]. The main multisensor core-logger
(MSCL) measurements obtained from the unsplit cores include gamma
ray absorption bulk density (GRA), low-field (0.1 mT) volume
magnetic susceptibility (MS) using two loops with offset
frequencies of 621 Hz and 513 Hz with measurements adjusted to the
standard 565 Hz response [Backman et al., 2006], compressional wave
velocity (PWL) and natural gamma ray (NGR) emissions [Backman et
al., 2006; Blum, 1997]. The sample resolu- tion ranges between 2 to
5 cm between samples. For color reflectance (L*, b*) [Blum, 1997]
the sampling interval is every 5 cm. Technically, we make use of
the ‘‘spliced’’ section, however, in reality there is almost no
overlap between cores below the top few tens of meters. We use the
meters composite depth (mcd) scale [Backman et al., 2006] to assign
different nonoverlapping depths to different cores. A revised
composite depth scale is now available for the upper tens of meters
[O’Regan et al., 2008], which is important to interpret
beryllium-10 de- rived ages [Frank et al., 2008; Backman et al.,
2008], magnetostratigraphic interpretations and high-resolution
studies of Plio-Pleistocene sediments, but does not influ- ence our
analysis or interpretation.
2.2. X-Ray Fluorescence Measurements
[8] The multisensor track measurements obtained from the ACEX cores
allow assessment of cyclical patterns in sediment density,
mineralogical and grain size composition,
and color. However, the ACEX core quality was compro- mised by the
use of an extended core barrel–type recovery system [Backman et
al., 2006]. This system can result in partially filled core liners,
or biscuiting, problems that are generally avoided when using a
hydraulic piston coring system. A significant help in understanding
whether the MSCL data patterns are due to variations of sediment
properties, or due to drilling disturbances, is provided through
the high-resolution measurement of (relative) major element
fluctuations, measured on the surface of split archive halves. The
detailed description of measurement techniques, calibration of XRF
counts per seconds to absolute concentrations, and an evaluation of
correlation between the elements (K, Ca, Fe, Al, Si, S, Mn, Ti) is
given by Spofforth et al. [2008] (see also Sangiorgi et al.
[2008b]). These measurements are shown to be useful to ascertain
the relative contributions from terrigenous and biogenic matter to
the bulk sediment. [9] Spofforth et al. [2008] also generated a
calibration of
elemental concentrations to allow the conversion of rawXRF counts
to calibrated concentrations. Similar to previous work, this
analysis uses elemental concentration variations and physical
property measurements as a semiquantitative geochemical approach to
decipher sedimentary patterns [Wehausen and Brumsack, 1999; Lourens
et al., 2001; Palike et al., 2001]. The sampling resolution for
calibra- tion measurements (via ICP-AES) from subsections of the
record was chosen on the basis of the shipboard age model [Backman
et al., 2006], aiming for a resolution in time of about 1 ka, and
resulting in actual sampling intervals of between 0.5 cm and 2 cm
[Spofforth et al., 2008].
2.3. Other Measurements and Proxies
[10] Owing to the often noisy nature of the MSCL measurements
obtained from the short XCB ACEX cores, it has proven difficult to
obtain meaningful estimates for sedimentary patterns from just one
single parameter. For this reason one of the approaches we chose
here is to consider multiple proxy measurements simultaneously, us-
ing an algorithmic approach. Additional data sets that were
generated as part of the ACEX effort include measurements on bulk
organic carbon from the organic rich units 1/6 and 2, mostly from
the Paleogene [Stein et al., 2006] as well as isothermal remanent
magnetization (IRM) and anhysteretic remanent magnetization (ARM)
magnetic proxies produced for the Paleogene as part of the onshore
standard measure- ment program (J. Gattacceca, unpublished data,
2006). All magnetization measurements were performed on U-channel
samples (150 cm length, 2 cm by 2 cm cross section) using a 2G DC
SQUID magnetometer. Measurement intervals were 2 cm. The field
strength resolution is about 106 A m1. ARM was measured after
imparting a 100 mT field, superimposed with a 100 mT alternating
field. IRM was measured after imparting a field of 1000 mT. The
full data set also includes field intensities after demagnetization
with alternating fields of 50 mT (peak value). Magnetic proxy data
(ARM/IRM ratio) were considered separately by O’Regan et al. [2008]
for an evaluation of the Quaternary part of the ACEX record.
Summary Figure 1 includes the
PA1S10 PALIKE ET AL.: ORBITAL TIMESCALES FROM THE ARCTIC
2 of 13
PA1S10
Figure 1. Summary plot of data used. (left) Meters composite depths
(mcd) and cores used in splice together with the (right)
sedimentation rates that result from the age model of Backman et
al. [2008] (with lithological units [Backman et al., 2006]). Data
plotted are XRF-derived elemental concentrations of Ti, Fe, Si, K,
and Al (in 1000 counts per second (cps)); organic carbon contents
(TOC) [Stein et al., 2006]; log isothermal remanent magnetization
(IRM) at 1000 mT (in A m1); and multisensor track properties
electrical resistivity (RES) (W m), natural gamma ray (NGR) (gamma
counts s1), compressional wave velocity (PWL) (sound velocity, km
s1), b* and L* (color reflectance, percent), gamma ray absorption
(GRA) (bulk density, g cm3), and volume specific magnetic
susceptibility (log10(instrument units +106
International System of Units)). Grey shaded intervals labeled a,
b, c, and d represent snapshot intervals analyzed in detail in this
study.
PA1S10 PALIKE ET AL.: ORBITAL TIMESCALES FROM THE ARCTIC
3 of 13
PA1S10
sedimentation rates predicted using the age model of Backman et al.
[2008].
2.4. Data Cleaning Procedures
[11] A first and necessary step before attempting frequency
analysis of time series from sediment cores often involves the
preparation and pruning of existing data sets, so as to avoid
spurious spikes in the records and to enhance the signal-to-noise
ratio prior to analysis. Outliers originating from end-of-core
effects (e.g., end caps, partially filled core liner at the end of
cores or sections) were readily identified visually, and typically
were removed prior to spectral analysis. For some measurements
(e.g., IRM) we applied an algorithmic spike reduction, by filtering
out all those points that lie outside the 1.5 one-sided standard
deviation from the median over 5 running points. This step was
successful in eliminating anomalous spikes in the records. In the
case of the ACEX cores, several spikes, particularly in the bulk
density record, were not the result of core disturbances, but
instead resulted from the occurrence of both drop stones and pyrite
nodules in the records. While drop stones and pyrite nodule
occurrence may be partially driven by climatic variations, they do
not appear to linearly scale with a climatic forcing signal, and
are thus not analyzed here. We attempted to manually remove such
spikes prior to analysis. All records were resampled to an equal
depth step (while preserving the highest sample resolution within a
given record), linearly detrended, and normalized by subtracting
the arithmetic mean and dividing by the standard deviation [Weedon,
2003].
2.5. Frequency Ratio Snapshots
[12] In order to establish whether the ACEX physical property and
XRF records are compatible with a hypothesis that invokes orbital
forcing to explain the cyclical patterns observed, several
analytical approaches are possible. One approach we use is to take
subsections from the entire data set, spanning one or two cores
with the highest-quality measurements, and compute an average power
spectrum, using the multitaper method [Ghil et al., 2002].
Calculated spectra then enable us to apply a frequency ratio test
between significant peaks. Here we do not match the core data to
astronomical templates using available biostratigra- phy and
magnetostratigraphy, but instead analyze the fre- quency ratio of
the cycles that are preserved and assume that the ratio of dominant
sedimentary cycle frequencies should follow the pattern predicted
by astronomy (roughly 400 ka: 120 ka: 41 ka: 22 ka ratios). This
method, which is suitable for uncertain age control, has been
applied previously [Hilgen, 1991; Dinares-Turell et al., 2003;
Raffi et al., 2005], and allows a rough estimate of likely
sedimentation rates if Milankovitch frequencies at the predicted
ratio are present in the snapshots of cores analyzed.
2.6. Wavelet Analysis
[13] We run time-evolutive spectral analyses in the depth domain to
establish the presence of orbital signals [Weedon, 2003]. We apply
a continuous wavelet transform [Torrence and Compo, 1998] (online
software available from http://
atoc.colorado.edu/research/wavelets/, accessed 29 April 2007) in
order to establish how the spectral power of proxy
series vary with depth. This method requires interpolation of our
series to equal depth steps, which were chosen to preserve the
maximum depth resolution across the records (typically 2 cm
intervals). We show here the results from the entire stratigraphic
interval recovered in the case of the physical property data (0 to
404.68 mcd) (Figure 2), and over a short interval for units 1/6 and
2 for selected XRF measurements [Spofforth et al., 2008], organic
carbon contents [Stein et al., 2006], and IRM (Figure 3). [14] We
converted the astronomically calculated [Laskar
et al., 2004] insolation at 85N for June–July from an absolute
timescale to the composite depth scale (mcd), using a reverse
linear interpolation from the age model of Backman et al. [2008].
This allowed a comparison of predicted orbital cycle frequencies in
the depth domain to those obtained from the proxy series (Figure
2). We note that insolation quantities are typically dominated by
obliquity and preces- sion, and that geological records often show
a red back- ground noise, muting the higher frequencies [Weedon,
2003; Palike et al., 2006a]. We also calculate the predicted
position of the four main Milankovitch bands (405, 110, 41, 22 ka)
using the simple linear age model of Backman et al. [2008]. We use
the predictions from this linear age model as a visual aid to
establish where in the frequency-depth continuum one would expect
to see orbitally controlled cyclical patterns. This approach made
the double assump- tion that the sedimentation rates of Backman et
al. [2008] are correct and that records are potentially
astronomically forced.
2.7. Multichannel Singular Spectrum Analysis
[15] If several proxy measurements exist for the same depth
interval, it is necessary to reconcile the different information
contained in each series. The availability of multiple data sets
should allow the enhancement of the signal-to-noise ratio of a
common encoded signal. Each proxy curve is a mixture of a
climatological signal, noise, and nonperiod changes that are common
to all data series because of local geological conditions.
‘‘Noise’’ in proxy curves reflects measurement, calibration, or
sedimento- logical uncertainties. It is unpractical to analyze all
data series separately, and thus a useful approach is to apply
analysis to a combination curve of all measurements available. [16]
To achieve a decomposition into correlated signals
and noncorrelated signals (noise), we have applied an approach
previously developed for meteorological data sets, namely singular
spectral analysis (SSA) decomposition [Vautard and Ghil, 1989;
Vautard et al., 1992]. Here we used an extended method
(multichannel singular spectrum analysis (MSSA)) that can be
applied to multiple proxy channels [Plaut and Vautard, 1994; Jiang
et al., 1995]. This approach allows us, without a priori knowledge,
(1) to extract a subset of oscillations that is common to all proxy
series considered, (2) to establish which are the dominant
periodicities, and (3) to reconstruct the higher-order princi- pal
components identified in item 2, thus enhancing the signal-to-noise
ratio. We ran several experiments using the MSSA approach to first
identify those components that show correlated oscillations for
depth intervals labeled
PA1S10 PALIKE ET AL.: ORBITAL TIMESCALES FROM THE ARCTIC
4 of 13
PA1S10
‘‘b’’ and ‘‘c’’ in Figure 1, and then reconstructing selected
principal components that show strong and stable common
oscillations between the proxies considered.
3. Results
[17] In section 3.1 we present sedimentation rates for snapshot
depth intervals from the ACEX record, indicated by a–d in Figure 1.
We apply this strategy because of the large number of core gaps
across the record, which make a continuous analysis not possible.
We compared the derived
sedimentation rate from each snapshot interval with the integrated
age model of Backman et al. [2008].
3.1. Pleistocene Sedimentation Rates: 0–27 mcd
[18] A detailed analysis of rock magnetic properties and physical
proxies resulted in a new high-resolution age model for the
Pleistocene ACEX section, and is presented fully by O’Regan et al.
[2008]. The depth interval covered in that study is indicated by
‘‘a’’ in Figure 1. O’Regan et al. [2008] exploited characteristic
patterns in the proxy data to allow correlation with existing
records and ages. They also
Figure 2. Continuous wavelet transform (CWT) analysis [Torrence and
Compo, 1998] of data in depth domain (entire record). (a) Ages and
apparent sedimentation rates of Backman et al. [2008]. Note major
hiatus between 18 and 44 Ma [Sangiorgi et al., 2008a]. (b) CWT
analysis of insolation calculated at 85N for June–July [Laskar et
al., 2004] after remapping to depth using age model from Figure 2a.
Superimposed are white shaded bands that indicate the predicted
position of 22, 41, 110, and 405 ka Milankovitch bands. (c–e) CWT
analyzed for physical property data bulk density (GRA), lightness
(L*), and yellow-blue (b*) [Backman et al., 2006]. Sedimentary
units and meters composite depths (mcd) are indicated below. Curved
white edges indicate typical shape of ‘‘cone of influence,’’ where
edge effects influence the spectra. Similar shapes would extend
around each core gap (not plotted).
PA1S10 PALIKE ET AL.: ORBITAL TIMESCALES FROM THE ARCTIC
5 of 13
PA1S10
used a cyclostratigraphic approach to constrain paleomag- netic
interpretations back to the Cobb Mountain event (1200 ka). The new
high-resolution age model presented by O’Regan et al. [2008]
results in sedimentation rates that are very similar to the average
Neogene rates of about 1.45 ± 0.1 cm ka1 obtained by 10Be/9Be
analysis of Frank et al. [2008]. [19] Generally, the ages of
O’Regan et al. [2008] confirm
the overall high sedimentation rates around 10–20 m Ma1
revised recently for the Arctic Ocean Basin [Backman et al., 2004].
The wavelet spectra calculated in the depth domain for color
reflectance (Figure 2e) confirm this finding by displaying
consistent power around the periods predicted by the 10Be/9Be age
model (periodicities around 0.25 m, 70 cm, and 1.5 m, corresponding
to climatic precession, obliquity, and the short eccentricity
frequency ratios). The continuous
wavelet transform analysis of physical properties (Figure 2)
indicates that sedimentation rates estimated from fre- quency
ratios of sedimentary cycles remain largely compat- ible with the
10Be/9Be–derived age model, but possibly indicating marginally
smaller sedimentation rates around 100 mcd.
3.2. Paleogene Sedimentation Rates: 200–404.8 mcd, 44–56.2 Ma
[20] The following three snapshots were selected from lithological
units 1/6 and 2 from the ACEX record, below a major hiatus
separating the Miocene from the middle Eocene at 198.7 mcd
[Sangiorgi et al., 2008a; Backman et al., 2008]. We note that this
hiatus itself is well charac- terized in depth by the XRF
measurements as a sharp break in elemental concentrations, with
elevated counts just above this major >26 Ma lasting hiatus
(Figure 3). 3.2.1. Sedimentary Unit 1/6 (Middle Eocene) [21]
Sedimentary units of the ACEX stratigraphic section
discussed here, and shown in Figure 1, are defined by Backman et
al. [2006]. Sedimentary unit 1/6 represents the youngest Paleogene
material that was recovered during ACEX, just below a major hiatus
[Sangiorgi et al., 2008a]. The wavelet analysis of this interval
(Figures 2 and 3) indicates that several sedimentation rates result
in cycle periods that show frequency ratios compatible with orbital
forcing. These sedimentation rates were variable across this unit.
For the interval between 203 and 220 mcd within unit 1/6 (marked as
‘‘b’’ in Figure 1), the physical properties, the XRF-derived
elemental concentrations and calibrations [Spofforth et al., 2008],
the organic carbon content [Stein et al., 2006] and the isothermal
remanent magnetization (IRM) all show a pronounced cyclicity. The
cycle period ranges from a short 15 cm spacing to longer decimeter
and meter scales (Figures 3 and 4), with strong correlations
between the different parameters. We make use of the MSSA
decomposition technique to isolate the largest amplitude
oscillations that are common to all proxies. Figure 4 shows an
example of this method, where the raw records are input into the
MSSA algorithm, computed with 200 lags for an interpolated sample
spacing of 0.5 cm, and computing 40 principal components. The
strongest and most coherent oscillations are then selected, and
used to reconstruct a new signal with a higher signal-to-noise
ratio. This recon- structed and ‘‘cleaned’’ signal explains about
42% of the total variance across all XRF parameters. We investigate
the spectral characteristics of individual reconstructed components
in more detail (Figure 5). The amplitude normalized spectral peaks
of individual reconstructed com- ponents occur at frequencies at
around 2–4 cycles m1 and 6–8 cycles m1. Some components show peaks
at both frequency ranges (e.g., principal reconstructed components
(PCRCs) 12 and 13), and band-pass filtering of the raw series
indicate the higher-frequency oscillations are not harmonics, but
real oscillations. [22] Figure 6a shows the XRF results from this
interval in
detail, and illustrates the observation that all records corre-
late extremely well with each other within this unit (decreases in
Fe correspond to increases in Ti, Al, Si). The cycle spacing in
this interval is also variable: between
Figure 3. CWT analyses as in Figure 2 but for depth interval
190–300 mcd (bottom of sedimentary unit 1/4 through base of unit 2)
for (a) XRF-derived Ti counts, (b) K counts, (c) organic carbon
TOC, and (d) log(IRM).
PA1S10 PALIKE ET AL.: ORBITAL TIMESCALES FROM THE ARCTIC
6 of 13
PA1S10
Figure 4. Unit 1/6 analysis, shown in Figure 1 as ‘‘b.’’ (top)
XRF-derived Ti, Al, and Si counts, normalized to standard
deviation, with multichannel singular spectrum analysis (MSSA)
reconstructions from the reconstructed principal components shown
below. There is one reconstructed principal component per element
(Ti, solid; Al, dashed; and Si, dotted). Grey components were not
used for reconstruction (they do not pass the paired frequency test
[Ghil et al., 2002]). Small-amplitude principal reconstructed
components (PCRCs) 21–40 were also not used or plotted. The
reconstructed components of the proxy series combined from Al, Si,
Fe, and Ti explain about 42% of the total variance. (bottom)
Reconstructed Ti for each reconstructed principal component for
comparison (orange).
PA1S10 PALIKE ET AL.: ORBITAL TIMESCALES FROM THE ARCTIC
7 of 13
PA1S10
206 and 208 mcd cycles are more closely spaced. From 210 to 216 mcd
their amplitude is reduced, and explained by a core gap. Between
216 and 220 mcd, the amplitude of individual cycles is higher than
in other intervals. [23] A Blackman-Tukey spectral analysis [Tukey,
1977] of
the signal reconstruction from this interval (Figure 6b) identified
several spectral peaks, where the high-frequency component
oscillates between 6 and 8 cycles per meter (15 cm spacing), and
the lower-frequency components are split with periods around 33 cm,
70 cm and 2 m. Figure 6b illustrates the frequency ratio approach
used to identify two possible sedimentation rate hypotheses. We
make use of a template of predicted astronomical frequencies (in
cycles per meter) as a function of sedimentation rate for climatic
precession (22 ka), obliquity (41 ka), and short (110 ka) and long
(405 ka) eccentricity. Identifying the 33 cm spaced oscillations as
climatic precession frequencies yields model 1, with sedimentation
rates between 10 and 20 m Ma1, which is compatible with the
position of the lower frequen- cy peaks identified. Alternatively,
if the 15-cm scale cycles are interpreted as precession-driven, a
different hypothesis (model 2) is possible, with resulting
sedimentation rates of
Figure 5. Spectral analysis of MSSA reconstruction principal
components. (a) Amplitude-sorted eigenvalues for each PCRC of
Figure 4 on a logarithmic amplitude scale. (b) Maximum entropy
spectrum [Press et al., 1992] for each PCRC of Figure 4.
Figure 6. Unit 1/6 analysis, shown as ‘‘b’’ in Figure 1. (a) XRF
data normalized to standard deviation and MSSA reconstruction of
significant principal components. Al, Si, and Ti anticorrelate with
Fe. (b) Blackman-Tukey spectral analysis of reconstruction from
Figure 6a (bottom plot, with 95% confidence interval and bandwidth
indicated), with two interpretations of frequency ratios labeled 1
and 2 if identifying spectral peaks with predicted position of
Milankovitch bands as a function of sedimentation rate (top plot).
(c) Relative timescale developed for Ti record using frequency
ratio labeled 1 in Figure 6b. (d) Wavelet analysis of Ti combined
cycles against depth, with superimposed Milankovitch bands using
age model from Figure 6c.
PA1S10 PALIKE ET AL.: ORBITAL TIMESCALES FROM THE ARCTIC
8 of 13
PA1S10
around 6 to 8 m Ma1. We note here that there is abundant pyrite
present within this unit, possibly influencing the high-frequency,
high-amplitude oscillations. At present, we cannot make a definite
choice as to whether model 1 or 2 is the correct one, based on
spectral arguments alone. How- ever, we prefer model 1, which is
compatible with other sedimentation rate constraints [Backman et
al., 2008]. However, hypothesis 1 implies that the higher-frequency
component with a cycle spacing of 15 cm represents half- precession
cycles, which is not likely given that these are predicted to occur
near the tropics only [Berger and Loutre, 1997]. The MSSA
decomposition shows that the high-frequency oscillations of the XRF
records occur in the upper (204–208 mcd) and lower intervals (217–
220 mcd) of lithological subunit 1/6. As both sedimentation rate
models 1 and 2 are compatible with frequency ratios predicted by
astronomy (with frequency ratios of 3:1.5:0.5 cycles m1 for model
1, or 7:3:1.5 for model 2), it is not possible, at present, to come
to a definite conclusion. One possibility that reconciles both
models are variations of sedimentation rate within the interval
studied by a factor of 2. This factor is common in geological
records [e.g., Palike et al., 2006b], and are also supported by a
good match between insolation target and data after a tentative
tuning of the records to insolation calculations (Figure 6c). [24]
Figure 6c shows how the sedimentation rate model 1
can be used to derive a relative (‘‘floating’’) age scale by
comparison of the XRF elemental variations with computed insolation
curves from Laskar et al. [2004]. Also shown are sedimentation
rates from this rudimentary ‘‘tuning,’’ again yielding linear
sedimentation rates of between 10 and 25 m Ma1. These show a visual
match between predicted Milankovitch bands and high-amplitude
‘‘ridges’’ in the wavelet spectrum (Figure 6d). Hence, for unit 1/6
we find that the biostratigraphic age control is compatible with
the cycle spacing observed, but there is also a different
possible
sedimentation rate model that is 50% slower than that predicted
from existing age control [Backman et al., 2008]. 3.2.2.
Sedimentary Unit 2 (Eocene), Core 2A-55X [25] We now extend our
analysis to an interval from
sedimentary unit 2, which is also characterized by high- amplitude
and coherent oscillations in XRF and physical properties, and for
which an additional set of biological proxies were generated
[Sangiorgi et al., 2008b]. This interval is marked by ‘‘c’’ in
Figure 1 and encompasses core 2A-55X, from about 236 to 241 mcd.
[26] A detailed analysis of available measurements from
this core is presented by Sangiorgi et al. [2008b]. The wavelet
analysis in Figure 3 suggests that the sedimentation rates of about
20 m Ma1 derived for Core 2A-55X within unit 2 can be extrapolated
downward to at least 270 mcd, based on the horizontal ‘‘ridges’’ in
the wavelet spectra that illustrate the presence of oscillations at
the main periods around 50 cm and 1 m. [27] Applying a light
‘‘tuning’’ to the records from Core
2A-55X and calculating a multitaper power spectrum from this tuned
record (Figure 7) [see also Sangiorgi et al., 2008b, Figure 9]
illustrates that the amplitude ratio of the obliquity and climatic
precession signal in the XRF and physical properties is about 1:1,
which has important ramifications for insolation models,
illustrated in section 4. 3.2.3. Sedimentary Unit 2 (Eocene), Core
4A-11X [28] The interval around 300 mcd was very sparsely
recovered during ACEX, but yielded a relatively complete Core
4A-11X which showed a remarkable record of Azolla fern spores
within unit 2 [Brinkhuis et al., 2006]. [29] Wavelet analysis for
this interval (Figures 2 and 3)
indicates that the sedimentation rate in this crucial interval was
also near 20 m Ma1, and that the predicted decrease in
sedimentation rate down core [Backman et al., 2008] in most
likelihood occurred below core 302-4A-11X.
4. Discussion
4.1. Sea Ice and Orbital Insolation
[30] Results from analyzing the Eocene ACEX time series needs wider
consideration of likely effects of orbital forcing on climate
processes in the high latitudes. The varia- tion of elemental
concentrations [Backman et al., 2006; Spofforth et al., 2008] and
biological proxies [Sangiorgi et al., 2008b] suggests a potential
role of sea ice variations in response to orbital forcing. The
presence of sea ice in the Arctic plays an important role in
interactions between insolation forcing and climatic response, for
example, through the albedo-temperature positive feedback. It also
controls exchange of heat between the ocean and the atmo- sphere,
and is implicated in variations of the thermohaline circulation
intensity. [31] Recently several modeling studies [Gallimore
and
Kutzbach, 1995; Jackson and Broccoli, 2003; Tuenter et al., 2005]
have investigated the response and role of sea ice in the Northern
Hemisphere in response to orbital insolation changes. DeConto et
al. [2007] investigated the feedback of sea ice on climate on
Antarctica. The high latitudes expe- rience large-amplitude
variations in orbital insolation on annual and Milankovitch
timescales (tens of thousands to
Figure 7. Interval ‘‘c’’ in Figure 1, modified from Sangiorgi et
al. [2008b]. Multitaper spectral analysis of insolation, GRA bulk
density, and summary signal extracted by MSSA from combined bulk
density, magnetic suscept- ibility, XRF-derived Ti, Al, and K, and
natural gamma measurements.
PA1S10 PALIKE ET AL.: ORBITAL TIMESCALES FROM THE ARCTIC
9 of 13
PA1S10
hundreds of thousands of years). A data set from the Arctic can
provide a first test of the predictions arising from these modeling
studies. [32] Gallimore and Kutzbach [1995] used a
low-resolution
general circulationmodel with extreme orbital configurations to
investigate how annual mean sea ice thickness and autumnal sea ice
coverage change under different orbital configurations. They found
that an increase in Earth’s obliquity (tilt) by three degrees
results in a decrease of mean sea ice thickness of 50%, and a
reduced autumnal sea ice coverage by 41%. Under a different forcing
experiment, using a relatively high eccentricity, and changing the
time of Earth’s closest approach to the sun (perihelion) from
North- ern Hemisphere (NH) summer to NH winter resulted in a 21%
reduction in mean ice thickness, but no change in the autumnal sea
ice coverage. The authors attribute the effect of larger obliquity
on sea ice to enhanced annual mean radiation, and the precessional
effect on seasonal interaction between those modeled processes that
govern sea ice growth and melt, and the heat transfer through sea
ice. For precession, a change from the date of perihelion from
summer to winter results in a reduction of summer radiation in the
northern extratropics of about 80 W m2, while for winter the
changes are smaller because of the overall smaller radiation, and
only reach about 20 W m2 at 45N latitude. In the case of the
increased tilt, and combined increased tilt and summer perihelion
experiments, the an- nual mean surface temperatures in the model
study were almost 6C higher over the high-latitude ocean area (60–
90N), resulting in less sea ice cover. The authors also found a
pronounced sensitivity of snow cover on the northern continents in
response to changed orbital configurations. Gallimore and Kutzbach
[1995] investigated the sea ice and snow response to insolation
changes using extreme values for the past 115 ka. They acknowledge
that their sea ice formulation ignores the effects of ice dynamics,
leads, multilayer heat conduction, brine pockets, and snow accu-
mulation on top of the ice, as well as the level of atmospheric CO2
concentrations, which are likely to have been very different for
the Eocene time slices [Pagani et al., 2005] we have analyzed from
the ACEX records. [33] In addition to the modeling result that
obliquity and
precession changes both have the potential to profoundly affect sea
ice coverage and extent in the Northern Hemi- sphere, largely
through changes in boreal summer insolation [Gallimore and
Kutzbach, 1995; Jackson and Broccoli, 2003; Tuenter et al., 2005]
showed that an additional effect during high summer insolation is a
reduction in the occur- rence of rapid (sub-Milankovitch timescale)
shifts in sea ice cover, as well as changes in the Atlantic
thermohaline circulation. [34] In contrast to the Northern
Hemisphere insolation
time series, which would predict a very strong forcing of
high-latitude climate by climatic precession, climate proxy records
show a large response of the global climate system on obliquity
timescales, at least during the late Pliocene and early Pleistocene
[e.g., Tiedemann et al., 1994]. More recently, a new theory for
insolation calculations was suggested [Huybers, 2006] to explain
these observations from climate records. This theory uses
integrated positive
degree days instead of mean monthly insolation time series,
resulting in time series that are dominated by obliquity instead of
climatic precession. The positive degree days time series is
clipped above a minimum threshold. The effect of this clipping is
that with higher thresholds the obliquity amplitude in these new
insolation records is reduced, and the amplitude of climatic
precession is en- hanced. It is thus possible to use observed
ratios of obliquity to climatic precession amplitudes from climate
records to obtain best fitting threshold values for positive degree
day insolation calculations.
4.2. Insolation Forcing in the Arctic: ACEX Data
[35] It is interesting to note that despite poor core recovery and
lack of overlapping cores that would allow the construction of a
detailed ‘‘splice’’ the method applied here can still result in
useful determinations of sedimentation rates, and constrain rates
of change within the Cenozoic record from the Arctic Ocean. It was
useful to analyze a large number of different proxy measurements,
as any individual data series has a too low signal-to-noise ratio.
Investigating several ‘‘snapshots’’ of Eocene time, the revised
picture of sedimentation history within the Arctic Ocean [Backman
et al., 2004] demonstrates that it was not sediment starved
throughout the Paleogene. [36] Our analysis also allows us to
provide information on
dominant orbital forcing parameters on climate proxy records during
the Eocene, specifically, the relative ampli- tude of obliquity
versus climatic precession response. The correlation of proxy
parameters in the Eocene part of the ACEX cores strongly suggests
that higher elemental con- centrations of Al, Ti and K (as derived
by XRF), for example, are controlled by sea ice transport
[Sangiorgi et al., 2008b]. The identification of orbitally driven
sedimen- tary cycles close to the geographic north pole is
interesting, as it would allow the testing of several paradigm
views that arose from the original Milankovitch theory. [37] In
particular there is still a large debate over why the
orbital cycles recorded in ice and marine cores are not dominated
by 41 ka and 21 ka cycles during the last several hundred thousand
years, as insolation forcing would predict. Instead, records show
100 ka spaced oscillations that have traditionally been explained
as eccentricity cycles. However, several hypotheses have been put
forward that explain the 100 ka spacing as a set of noncomplete
obliquity cycles, that stochastically appear every 80 or 120 ka
[Huybers and Wunsch, 2005]. [38] Our records from ACEX allow us to
explore the new
prediction of Huybers [2006] and previous modeling studies [Tuenter
et al., 2005] by analyzing the relative amplitude ratio between
obliquity and climatic precession cycles in our data. The model of
Huybers [2006] allows the calcula- tion of his new insolation time
series, for a given latitude and for a set of cutoff insolation
values. Figure 8a shows the predicted ratio of obliquity to
precession amplitude in the local insolation forcing. Obliquity is
by far overpowering the precession component in the local forcing
for cutoff values higher than about 100 W m2. [39] Huybers [2006]
used a very strong global correlation
between observed local air temperature, and local instanta-
PA1S10 PALIKE ET AL.: ORBITAL TIMESCALES FROM THE ARCTIC
10 of 13
PA1S10
neous insolation lagged by 30 d, to make a prediction of what this
cutoff point might be globally, and arrives at a value of about 250
W m2, resulting in a much reduced precession component in local
forcing. [40] In the case of the ACEX location, near the
geographic
north pole (87.5N), and within the Arctic circle with constant
daylight during summer and constant darkness during winter, we
could expect a strong local forcing of sea ice formation, melting
and transport in response to the local insolation forcing. [41] We
have applied the method of Huybers [2006] to
investigate the predicted response of local temperature to
insolation by comparing the temperature records from WMO weather
station 04301 (Kap Morris Jesup, north Greenland, 83390N, 033220W),
with insolation computed at 85N (Figure 8). Figure 8a shows a
strong correlation between calculated local insolation (lagged by
26 d, which provides the best fit) and observed daily temperatures.
Figure 8b illustrates that the cutoff insolation value for average
daily temperatures above 0C is about 550 W m2. In Figure 8c, this
cutoff value would predict a local insolation forcing that is
dominated by precession with an amplitude ratio of obliquity versus
climatic precession of about 1:10, which might be taken as the
temperature most sensitive for the melting of sea ice. [42] In the
ACEX records, however, we establish an
amplitude ratio closer to 1 (Figure 7). While these results
indicate that the amplitude ratio of obliquity to climatic
precession, predicted from the present-day transfer function of
local insolation to temperature, does not directly apply to the
Eocene time series, we nevertheless find that the methodology used
by Huybers [2006] would predict cutoff insolation values for the
Arctic (550 W m2) that are not very different to those suggested by
the amplitude spectra calculated from the ACEX values (450–500 W
m2). We therefore conclude that the methodology of Huybers [2006]
warrants further investigation, and comparison with data records
from other latitudes and time slices. Our results are also
compatible with the influence of obliquity and climatic precession
established through modeling studies for the past glacial cycle
[Gallimore and Kutzbach, 1995; Jackson and Broccoli, 2003; Tuenter
et al., 2005].
5. Summary and Conclusions
[43] In summary, we evaluate the nature of oscillations in physical
proxy measurements and XRF-derived elemental concentrations from
the Arctic Ocean during snapshots of the Cenozoic. Even though the
stratigraphic recovery is incomplete, advanced time series analysis
methods allow us to put additional constraints on sedimentation
rates during the Eocene. The determined sedimentation rate options
are always compatible with those predicted by lower-resolution
biostratigraphic control, and yield sedimentation rates of the
order of 10–20 m Ma1. We establish that recent revised methods of
insolation calculations are useful to make quantitative comparisons
between the data recovered during ACEX, and predicted behavior of
insolation. The ACEX records show a ratio of orbital obliquity and
climatic precession imprint that is close to 1:1, and would
predict
Figure 8. (a) Average and maximum daily temperature at weather
station 04301 (84N), plotted from 1 January and averaged from 20
years of observation against local daily insolation, lagged by 26
calendar days. (b) Crossplot of lagged insolation and daily
temperature from Figure 8a. (c) Amplitude ratio of
obliquity:climatic precession from Huybers [2006] calculation as a
function of cutoff insolation value, with amplitude ratio of near
1:1 from ACEX record marked (see Figure 7).
PA1S10 PALIKE ET AL.: ORBITAL TIMESCALES FROM THE ARCTIC
11 of 13
PA1S10
an insolation cutoff value forHuybers [2006] positive degree day
insolation calculation of about 450–500 W m2, close to the local
very high-latitude value predicted theoretically (550 W m2).
[44] Acknowledgments. We are grateful to the expedition onshore and
offshore science party, as well as the crews of the I/B Oden, I/B
Vidar
Viking, and I/B Sovetsky Soyuz. This research used samples provided
by the Integrated Ocean Drilling Program (IODP). IODP is sponsored
by the U.S. National Science Foundation and participating countries
under the man- agement of Joint Oceanographic Institutions, Inc.
Financial support was provided by the Nuffield Foundation and the
U.K. Natural Environment Research Council (NERC) (reference
NE/D000343/1) to H.P. and by NERC studentship (reference
NER/S/A/2005/13474) to D.J.A.S.
References Backman, J., M. Jakobsson, R. Lovlie, L. Polyak, and L.
A. Febo (2004), Is the central Arctic Ocean a sediment starved
basin?, Quat. Sci. Rev., 23(1113), 1435 – 1454, doi:10.1016/
j.quascirev.2003.12.005.
Backman, J., K. Moran, D. B. McInroy, L. A. Mayer, and the
Expedition 302 Scientists (2006), Arctic Coring Expedition (ACEX),
Proc. Integr. Ocean Drill. Program, 302.
Backman, J., et al. (2008), Age model and core- seismic integration
for the Cenozoic Arctic Coring Expedition sediments from the Lomo-
nosov Ridge, Paleoceanography, doi:10.1029/ 2007PA001476, in
press.
Berger, A., and M. F. Loutre (1997), Intertropical latitudes and
precessional and half-precessional cycles, Science, 278, 1476–1478,
doi:10.1126/ science.278.5342.1476.
Billups, K., H. Palike, J. E. T. Channell, J. C. Zachos, and N. J.
Shackleton (2004), Astro- nomical calibration of the late Oligocene
through early Miocene geomagnetic polarity time scale, Earth
Planet. Sci. Lett., 224(12), 33–44,
doi:10.1016/j.epsl.2004.05.004.
Blum, P. (1997), Physical properties handbook: A guide to the
shipboard measurement of phy- sical properties of deep-sea cores,
Tech. Note 26, Ocean Drill. Program, College Station, Tex.
(Available at http://www-odp.tamu.edu/
publications/tnotes/tn26/INDEX.HTM).
Brinkhuis, H., et al. (2006), Episodic fresh surface waters in the
Eocene Arctic Ocean, Nature, 441, 606–609,
doi:10.1038/nature04692.
DeConto, R. M., D. Pollard, and D. Harwood (2007), Sea ice feedback
and Cenozoic evolu- tion of Antarctic climate and ice sheets, Pa-
leoceanography, 22, PA3214, doi:10.1029/ 2006PA001350.
Dinares-Turell, J., J. I. Baceta, V. Pujalte, X. Orue-Etxebarria,
G. Bernaola, and S. Lorito (2003), Untangling the Palaeocene
climatic rhythm: An astronomically calibrated early Pa- laeocene
magnetostratigraphy and biostratigra- phy at Zumaia (Basque basin,
northern Spain), Earth Planet. Sci. Lett., 216(4), 483 –500,
doi:10.1016/S0012-821X(03)00557-0.
Frank, M., J. Backman, M. Jakobsson, K. Moran, M. O’Regan, J. King,
B. A. Haley, P.W. Kubik, and D. Garbe-Schonberg (2008), Beryllium
isotopes in central Arctic Ocean sediments over the past 12.3
million years: Stratigraphic and paleoclimatic
implications,Paleoceanography, 23, PA1S02,
doi:10.1029/2007PA001478.
Gallimore, R. G., and J. E. Kutzbach (1995), Snow cover and sea ice
sensitivity to generic changes in Earth orbital parameters, J. Geo-
phys. Res., 100(D1), 1103–1120.
Ghil, M., et al. (2002), Advanced spectral meth- ods for climatic
time series, Rev. Geophys., 40(1), 1003,
doi:10.1029/2000RG000092.
Hilgen, F. J. (1991), Astronomical calibration of Gauss to Matuyama
sapropels in the Mediter- ranean and implication for the
geomagnetic
polarity time scale, Earth Planet. Sci. Lett., 104(2 – 4), 226 –
244, doi:10.1016/0012- 821X(91)90206-W.
Huybers, P. (2006), Early Pleistocene glacial cycles and the
integrated summer insolation forcing, Science, 313, 508–511,
doi:10.1126/ science.1125249.
Huybers, P., and C. Wunsch (2005), Obliquity pacing of the late
Pleistocene glacial termi- nations, Nature, 434, 491–494,
doi:10.1038/ nature03401.
Jackson, C. S., and A. J. Broccoli (2003), Orbi- tal forcing of
Arctic climate: Mechanisms of climate response and implications for
conti- nental glaciation, Clim. Dyn., 21, 539–557,
doi:10.1007/s00382-003-0351-3.
Jiang, N., J. D. Neelin, and M. Ghil (1995), Quasi-quadrennial and
quasi-biennial varia- bility in the equatorial Pacific, Clim. Dyn.,
12, 101–112.
Krylov, A. A., I. A. Andreeva, C. Vogt, J. Backman, V. V.
Krupskaya, G. E. Grikurov, K. Moran, and H. Shoji (2008), A shift
in heavy and clay mineral provenance indicates a middle Miocene
onset of a perennial sea ice cover in the Arctic Ocean,
Paleoceanography, doi:10.1029/2007PA001497, in press.
Kuiper, K. F., F. J. Hilgen, J. Steenbrink, and J. R. Wijbrans
(2004), 40Ar/39Ar ages of tephras inter- calated in astronomically
tuned Neogene sedi- men t a r y s e q u e n c e s i n t h e e a s t
e r n Mediterranean, Earth Planet. Sci. Lett., 222(2), 583–597,
doi:10.1016/j.epsl.2004.03.005.
Laskar, J. (1999), The limits of Earth orbital calculations for
geological time-scale use, Phi- los. Trans. R. Soc. London, Ser. A,
357(1757), 1735–1759, doi:10.1098/rsta.1999.0399.
Laskar, J., F. Joutel, and F. Boudin (1993), Orbi- tal,
precessional, and insolation quantities for the Earth from 20 Myr
to +10 Myr, Astron. Astrophys., 270, 522–533.
Laskar, J., P. Robutel, F. Joutel, M. Gastineau, A. Correia, and B.
Levrard (2004), A long term numerical solution for the insolation
quantities of the Earth, Astron. Astrophys., 428 , 2 6 1 – 285 , d
o i : 1 0 . 1 0 5 1 / 0 0 04 - 6361:20041335.
Lourens, L. J., R. Wehausen, and H. J. Brumsack (2001), Geological
constraints on tidal dissipa- tion and dynamical ellipticity of the
Earth over the past three million years, Nature, 409, 1029–1033,
doi:10.1038/35059062.
Lourens, L. J., F. J. Hilgen, J. Laskar, N. J. Shackleton, and D.
Wilson (2005), The Neo- gene period, in A Geological Time Scale
2004, edited by F. M. Gradstein et al., pp. 409–440, Cambrige Univ.
Press, New York.
Milankovitch, M. (1941), Kanon der Erdbes- trahlung und seine
Anwendung auf das Eiszei- tenproblem, Spec. Publ., vol. 132, R.
Serb. Sci., Belgrad. (English translation Israel Pro- gram for Sci.
Transl., Jerusalem, 1969).
Moran, K., et al. (2006), The Cenozoic pa- laeoenvironment of the
Arctic Ocean, Nature, 441, 601–605, doi:10.1038/nature04800.
O’Regan, M., J. W. King, J. Backman, M. Jakobsson, K.Moran, C.
Heil, T. Sakamoto, T. Cronin, and R. Jordan (2008), Constraints on
the Plio-Pleistocene chronology of sediments from the Lomonosov
Ridge, Paleoceanogra- phy, doi:10.1029/2007PA001551, in
press.
Pagani,M., J. C. Zachos, K. H. Freeman, B. Tipple, and S. Bohaty
(2005), Marked decline in atmo- spheric carbon dioxide
concentrations during the Paleogene, Science, 309, 600 – 603,
doi:10.1126/science.1110063.
Palike, H., and N. J. Shackleton (2000), Con- straints on
astronomical parameters from the geological record for the last 25
Myr, Earth Planet. Sci. Lett., 182(1), 1–14, doi:10.1016/
S0012-821X(00)00229-6.
Palike, H., N. J. Shackleton, and U. Rohl (2001), Astronomical
forcing in late Eocene marine sediments, Earth Planet. Sci. Lett.,
193(34), 5 8 9 – 6 0 2 , d o i : 1 0 . 1 0 1 6 / S 0 0 1 2 -
821X(01)00501-5.
Palike, H., J. Laskar, and N. J. Shackleton (2004), Geologic
constraints on the chaotic diffusion of the solar system, Geology,
32(11), 929–932, doi:10.1130/G20750.1.
Palike, H., R. D. Norris, J. O. Herrle, P. A. Wilson, H. K. Coxall,
C. H. Lear, N. J. Shackleton, A. K. Tripati, and B. S. Wade
(2006a), The heartbeat of the Oligocene climate system, Science,
314, 1894 – 1898, doi:10.1126/ science.1133822.
Palike, H., J. Frazier, and J. C. Zachos (2006b), Extended
orbitally forced palaeoclimatic re- cords from the equatorial
Atlantic Ceara Rise, Quat. Sci. Rev., 25(23 – 24), 3138 – 3149,
doi:10.1016/j.quascirev.2006.02.011.
Plaut, G., and R. Vautard (1994), Spells of low- frequency
oscillations and weather regimes in the Northern Hemisphere, J.
Atmos. Sci., 51 (2 ) , 210 – 236 , do i :10 .1175 /1520-
0469(1994)051<0210:SOLFOA>2.0.CO;2.
Press, W. H., S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery
(1992), Numerical Recipes in C: The Art of Scientific Computing,
2nd ed., pp. 656–706, Cambridge Univ. Press, Cam- bridge, U.
K.
Raffi, I., J. Backman, and H. Palike (2005), Changes in calcareous
nannofossil assem- blages across the Paleocene/Eocene transi- tion
from the paleo-equatorial Pacific Ocean, Palaeogeogr.
Palaeoclimatol. Pa- laeoecol., 226(12), 93 – 126, doi:10.1016/
j.palaeo.2005.05.006.
Raffi, I., J. Backman, E. Fornaciari, H. Palike, D. Rio, L.
Lourens, and F. Hilgen (2006), A review of calcareous nannofossil
astrobiochro- nology encompassing the past 25 million years, Quat.
Sci. Rev., 25, 3113 – 3137,
doi:10.1016/j.quascirev.2006.07.007.
PA1S10 PALIKE ET AL.: ORBITAL TIMESCALES FROM THE ARCTIC
12 of 13
PA1S10
Sangiorgi, F., H.-J. Brumsack, D. A. Willard, S. Schouten, C. E.
Stickley, M. O’Regan, G.-J. Reichart, J. S. Sinninghe Damste, and
H. Brinkhuis (2008a), A 26 million year gap in the central Arctic
record at the greenhouse- icehouse transition: Looking for clues,
Paleo- ceanography, 23, PA1S04, doi:10.1029/ 2007PA001477.
Sangiorgi, F., E. E. van Soelen, D. J. A. Spofforth, H. Plike, C.
E. Stickley, K. St. John, N. Ko, S. Schouten, J. S. Sinninghe
Damst, and H. Brinkhuis (2008b), Cyclicity in the mid- dle Eocene
central Arctic Ocean sediment record: Orbital forcing and
environmental response, Paleoceanography, doi:10.1029/
2007PA001487, in press.
Shackleton, N. J. (2000), The 100,000–year ice- age cycle
identified and found to lag tempera- ture, carbon dioxide, and
orbital eccentricity, Science, 289, 1897 – 1902, doi:10.1126/
science.289.5486.1897.
Spofforth, D. J. A., H. Palike, and D. Green (2008), Paleogene
record of elemental concen- trations in sediments from the Arctic
Ocean ob- tained by XRF analyses, Paleoceanography,
doi:10.1029/2007PA001489, in press.
Stein, R., B. Boucsein, and H. Meyer (2006), Anoxia and high
primary production in the Paleogene central Arctic Ocean: First
detailed records from Lomonosov Ridge, Geophys. Res. Let t . , 33 ,
L18606, doi :10.1029/ 2006GL026776.
Tiedemann, R., M. Sarnthein, and N. J. Shackleton (1994),
Astronomic timescale for the Pliocene Atlantic d18O and dust flux
records of Ocean Drilling Program Site 659, Paleoceanography, 9(4),
619–638.
Torrence, C., and G. P. Compo (1998), A prac- tical guide to
wavelet analysis, Bull. Am. Meteorol. Soc., 79(1), 61–78,
doi:10.1175/ 1520-0477(1998)079<0061:APGTWA>2.0. CO;2.
Tuenter, E., S. K. Weber, F. J. Hilgen, and L. J. Lourens (2005),
Sea-ice feedbacks on the cli- matic response to precession and
obliquity for- cing, Geophys. Res. Lett., 32, L24704,
doi:10.1029/2005GL024122.
Tukey, J. (1977), Exploratory Data Analysis, Addison-Wesley,
Reading, Mass.
Vautard, R., and M. Ghil (1989), Singular spectrum analysis in
nonlinear dynamics, with applications to paleoclimatic time series,
Physica D, 35, 395–424, doi:10.1016/0167- 2789(89)90077-8.
Vautard, R., P. Yiou, andM.Ghil (1992), Singular- spectrum
analysis: A toolkit for short, noisy chaotic signals, Physica D,
58, 95 – 126, doi:10.1016/0167-2789(92)90103-T.
Wade, B. S., and H. Palike (2004), Oligocene climate dynamics,
Paleoceanography, 19, PA4019, doi:10.1029/2004PA001042.
Weedon, G. P. (2003), Time-Series Analysis and Cyclostratigraphy,
Cambridge Univ. Press, New York.
Wehausen, R., and H. J. Brumsack (1999), Cyclic variations in the
chemical composition of east- ern Mediterranean Pliocene sediments:
A key for understanding sapropel formation, Mar. Geol., 153(1 – 4),
161 – 176, doi:10.1016/ S0025-3227(98)00083-8.
J. Gattacceca, Geophysics and Planetology,
CEREGE, CNRS, University of Aix-Marseille 3, F-13628
Aix-en-Provence, France. M. O’Regan, Graduate School of
Oceanogra-
phy, University of Rhode Island, Narragansett, RI 02882, USA. H.
Palike and D. J. A. Spofforth, National
Oceanography Centre, University of Southampton, European Way,
Southampton SO14 3ZH, UK. (
[email protected])
PA1S10 PALIKE ET AL.: ORBITAL TIMESCALES FROM THE ARCTIC
13 of 13